I'm not an engineer, nor even an Amateur, but I can start the answer in a simple fashion.

Essentially, the ratio of the ferrite rod length compared to its diameter determines the nulling quality. The larger the ratio, the better the nulling. A second minor factor is the length of the coil winding compared to the length of the ferrite rod. The effect is small, and generally a winding length of 1/3 to 1/2 of the ferrite rod length improves nulling a small bit by keeping the coil away from the ferrite rod cylinder-ends.

Sometimes one encounters a non-cylinder shape that is wider than thick. One can presume the circumference of the shape as an equivilent circle. Extreme cases exist, like so-called "Antenna Strips": some have widths 5 or 6 times the thickness. The reasoning is to present a large surface area to the signal received (a so-called apparent diameter). An easy test is to hold the radio antenna vertically, and tune in the station. The cylinder will not receive well, if at all. The ferrite antenna strip can be tuned due to the wide width facing perpendicular to the station.

When coupled with good rf circuitry, the front-end of a radio is greatly improved, preserving sensitivity, with a low noise-floor (that enhances the nulling effect).

Paul S. in CT